Molecular modeling methodologies (molecular dynamics, conformational searching, Monte Carlo) used data from the crystallized structure of bovine rhodopsin (excluding the intracellular and extracellular domains), which is the only mammalian 7-transmembrane receptor crystallized to date, in order to develop a model of the delta-opioid receptor by in silico methods; i.e., computer-directed mutagenesis to ensure that the sequence of the rhodopsin format coincided with that of the delta-opioid receptor by exchanging specific amino acids. A variety of delta agonists and antagonists based on the Dmt-Tic pharmacophore derived from X-ray diffraction analyses of three selective compounds with different specificities (delta- and mu-opioid receptor selective, and non-selective), as well as specific mu-opioid receptor agonists, which should have very low affinity with the delta-opioid receptor, were docked into the proposed binding pocket. The ligand-binding domain was initially determined from data on site-directed mutagenesis obtained from the literature. The minimized molecular models of the ligands reflected their known biological activities and receptor affinities and conformational changes in the peptides were initially examined by 1-H NMR (COSY, NOESY, HOHAHA, ROESY, DQF-COSY experiments), CD under varying solvent and temperature conditions. In terms of the ligands, the aromatic ring distance may be a singularly important characteristic which distinguishes delta-opioid receptor antagonists and agonists for both mu- and delta-opioid receptors providing a presumptive "receptor-bound conformation" in spite of the inherent flexibility of the peptide. As anticipated, mu-opioid receptor agonists exhibited a poor fit in the delta receptor pocket region, confirming the application of this methodology. The topographical features observed with the Dmt-Tic pharmacophore differentiate it from all other peptides and its interaction with select side-chains in the receptor pocket. The data suggest that the presumed receptor-bound conformation of the peptide ligand and receptor involves stacking between aromatic rings and hydrogen bonding and that mu-opioid agonists poorly interacted with those residues specific for delta ligands. Furthermore, there appeared to be two regions in which agonists and antagonists interact, only one of which is shared by these two types of compounds. Thus, intra-ring distance of delta-opioid receptor antagonists may portend biological differences due to its fit within its receptor. Peptide analogues with dual receptor binding characteristics or selectivity for the mu-opioid receptor equally assisted in the application of molecular modeling in a predictive mode. Thus, model of the delta receptor and our delta- and mu-opioid antagonist and agonist pharmacophores will serve as scaffolds in the design of new potent ligands.[unreadable] [unreadable] Based on pharmacophores developed by delta-opioid receptor analogues containing Dmt-Tic and several low energy modles of Dmt-Tic-Bid derivatives, pharmacophores were generated for virtual screening using LigandScout software. Furthermore, pharmacophores were obtained for morphine (mu agonist), Nalt44 and SNC-80 (delta agonists) to validate the pharmacophore screening procedure. The morphine pharmacophore produced more than 1,100 hits, whereas Nalt44 and SNC-80 each generated a single hit in a screen of the Derwent World Drug Index (WDI). Virtual screens of the Dmt-Tic pharmacophores identified 7 hits from WDI: while 4 of these retrieved up to 100 hits and identified seeral Dmt-Tic derivatives in our opioid database, 3 produced hits with features absent but required for opioid binding. Similarly, the same 4 pharmacophores were screened using the ChemDiverse database (ChemDiv) resulting in 3-900 hit, but most lacked "opioid-like" features. However, with modifications, some hits could serve as leads for opioid drug candidates. These methods offer an alternative approach to identify revelant pharmacophores for virtual screening when bioactive ligand conformations and the receptor binding site are unknown.